Alzheimer’s Disease Based On Deep Learning Algorithm Intelligent Diagnosis Research

Alzheimer’s disease is an irreversible neurodegenerative disease of the brain,with a long incubation period,and its clinical symptoms gradually deteriorate with the increase of time.The clinical symptoms include memory loss,amnesia,language ability degradation and irreversibility.At present,the main clinical diagnosis basis is to observe the patient’s magnetic resonance image,so this paper proposes to identify the medical image of Alzheimer’s patients based on the good performance and characteristics of deep learning algorithm,which has far-reaching significance for intelligent diagnosis of Alzheimer’s disease.This paper obtains data from ADNI database and uses deep learning algorithm to identify and classify patients’ magnetic resonance images.The main work is as follows:(1)Data processing.A large number of medical images of patients with different degrees are collected in ADNI database.Firstly,the image format is converted,and the high-dimensional image data is converted into low-dimensional images,and then irrelevant images are manually screened.Finally,in order to prevent the over-fitting phenomenon of the model,the images are enhanced.The main image enhancement methods include scaling,rotation and symmetrical transformation.(2)The structure and principle of deep learning algorithm and the popular network model architecture are introduced in detail.In order to better identify the nuclear magnetic images of Alzheimer’s disease,this paper proposes an intelligent diagnosis of Alzheimer’s disease based on deep learning algorithm,that is,using Res Net,VGG,Efficient Net,Vi T,Dense Net to extract and classify the features of the nuclear magnetic images,so as to achieve the final classification effect.Through comparative experiments,it can be found that Res Net has the best learning ability,and can make better use of the extracted features to achieve a higher classification effect.(3)In this paper,an unsupervised feature learning algorithm,IPCA,and con-volutional neural network algorithm are designed.Firstly,the unsupervised feature learning algorithm is used to preprocess the original high-dimensional data,and the features of the high-dimensional data are mapped to the low-dimensional data to realize feature extraction and data dimensionality reduction.Then,the output dimensionality reduction data is used as the input of convolutional neural network algorithm,and the processed medical image data is further calculated by combining the advantages of the two algorithms.After verification,the improved neural network algorithm improves the classification accuracy of Alzheimer’s disease.